Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Propagation phenomena in nonlocal reaction-diffusionequations: An overview of the recent developments
Jerome Coville
INRA Biostatistique et Processus Spatiaux (BioSP) Avignon -France
& Processus Spatiaux
Biostatistique
Workshop Integrodifference Equations in Ecology: 30 years andcounting , September 18–23, 2016, BIRS Center, Banff, Canada
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Outline of the Talk
Motivation
Homogeneous modelsNonlocal Reaction diffusion modelKnown ResultsWhat is new
Inside DynamicsPhilosophy
Heterogeneous modelThe periodic casegeneric nonlinearity
Other nonlocal equationinvasion by adaptation
Research IDEas
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
MotivationStudy of the epidemic of the poplar rust in the Durance Valley
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Epidemic Data: Halkett F.; Xhaard C. (INRA, UMR IAM)
Figure : Evolution of the epidemic intensity
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Phenotypical & Genotypical Data: Halkett F.; Xhaard C.
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Main Questions
1. How to explain the epidemic data ?
2. How to explain the genotypical and phenotypical data ?
3. Evaluate the gain of collecting genotypical/phenotypical data ?
4. How to model interactions between genotypes/phenotypes and thespace?
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
A continuous mathematical approach
Reaction-dispersion models:The population can be represented by a mean density u(t, x) which isdriven by:
∂u∂t (t, x) = M[u](t, x)︸ ︷︷ ︸ + f (t, x, u(t, x),K[u](t, x))︸ ︷︷ ︸ .
Time variation Movement Growth
Dispersion term M[u](t, x): Movement of the population (differencebetween individuals which come to location x and those which leavelocation x.)
Reaction term f : population growth rate. Depends on environmentalcharacteristics (carrying capacity of the environment) and the structure ofcompetition endures by the population K[u].
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Dispersal models
Diffusion model: local dispersal into adjacent habitat
M[u](t, x) = ∂2x u(t, x);
=⇒ Local dispersal operator linked to random walks.
Integro-differential model: non-local dispersal
M[u](t, x) =ˆR
J(|x − y|)(u(t, y) − u(t, x)
)dy;
Dispersal kernel: J(x − y) is the probability distribution of jumping fromlocation y to location x.
=⇒ Nonlocal dispersal operator. Can include long distance dispersal events.
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Example of kernel dispersal
Thin-tailed kernel Fat-tailed kernel
DefinitionˆR
J (x)eαx < ∞ for some α > 0.
Definition
for any α > 0, J (x) ≥ e−α|x| for large |x|.
x0
J(x) = Ce−α|x|
J(x) = Ce−|x|2 J(x) = 1[−L,L](x)
x0
J(x) = Ce−α|x|/(1+ln(1+|x|))
J(x) = Ce−√
|x|J(x) = (1 + |x|)−3
(Diekmann 1979, Thieme 1979, Schu-macher 1980, Weinberger 1982, Coville etal. 2008)
(Medlock and Kot 2003, Yagisita 2009)
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Modelling the growth
Population of size N (t, x) at time t and position x. The increase in size:
N (t + δt, x) − N (t, x) = (nb of birth-nb of death) during δt.
Birth rate b and death rate d.
u(t + δt, x) − u(t, x) = bu(t, x)δt − du(t, x)δt.
δt → 0 :
∂tu(t, x) = (b − d)u(t, x).
.
......
b and d encode the demographic processes considered Allee effect or Not Competition between the individuals or not
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Example of population growth in homogeneous media
f is a function, f (0) = f (1) = 0 and f ′(1) < 0.
(a) KPP (b) monostable non-degenerate (c) degenerated monostable
(d) ignition (e) bistable
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Answer to the first question
.The best model (C, Fabre, Halkett, Soubeyrand, Xhaard, preprint)..
......
After a statistical treatment of the data within the class of Fisher-KPP de-mographic function, the best model that fits the epidemic data is :
∂tu(t, x) = J ⋆ u(t, x) − u(t, x) + u(t, x)(r(x) − u(t, x)) (1)
with
J (z) ∼ e−|z|γ
with γ = 0.15
and
r(x) :=
K1 in (R0,+∞)K2 in (−R0,R0)K3 in (−∞,−R0)
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
A first conclusion
.
......Reaction diffusion equation with nonlocal dispersal are useful to explain realdata !!!
.To go further in the understanding of the data..
......
We need to investigate in more details The propagation phenomena for solution of the equation (1) The role of the nonlinearity in such propagation phenomena ways to evaluate the speed of propagation the role of the heterogeneity the role of the kernel
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
The Homogeneous Situation
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
The non local Reaction diffusion equation
∂tu = J ∗ u − u + f (u). (2)
J ≥ 0,´R J(z) dz = 1 and
´R J (z)|z| dz < +∞.
f is a smooth function, f (0) = f (1) = 0 and f ′(1) < 0.
(f) KPP (g) monostable non-degenerate (h) degeneratedmonostable
(i) ignition (j) bistable
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
The main questions to ask
Initial data is front-like: 0 ≤ u0 ≤ 1, lim infx→−∞ u0(x) > 0,
How does the stable state 1 invade or not the state 0?
Is there travelling fronts solution to (2)?
On what conditions on J and f such a solution exists
Can we characterise the spreading speed?
When such solution do not exist, what happens for the Cauchyproblem ?
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Known Results
.Travelling wave: (Bates-Fife-Ren-Wang (1997), Chen(1997), Alberti-Belletini(1998), Coville(2003,2006,2007),Yagisita (2009))..
......
Assume J is as above and f is a bistable or ignition nonlinearity, then thereexists a unique c such that there exists a travelling wave with speed c, i.e.there exists (φ, c) such that φ ∈ L∞, monotone and satisfying
ˆR
J (x − y)(φ(y) − φ(x))dy + cφ′(y) + f (φ(y)) = 0, in R
φ(−∞) = 1 and φ(+∞) = 0.
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Known Results.Travelling wave: (Schumacher 1980, Carr-Chmaj 2004, Coville et al.2003,2007,2008, Zhao et al. (2007,2008,2009), Yagisita 2009, ...)..
......
Assume J is a thin tailed kernel and f is a monostable nonlinearity, thenthere exists c∗ such that for all c ≥ c∗, there exists a travelling wave withspeed c, i.e. there exists (φ, c) such that φ ∈ L∞, monotone and satisfying
ˆR
J (x − y)(φ(y) − φ(x))dy + cφ′(y) + f (φ(y)) = 0, in R
φ(−∞) = 1 and φ(+∞) = 0.
(Lutcher et al., 2005 ): if u0 is compactly supported, the spreading speed c ofu satisfies c = c∗, the minimal speed of traveling fronts.
−100 −50 0 50 1000
0.5
1
x
u(t,x
)
Numerical obs: the solution converges to a traveling front with constant profile.
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Known Results
.Characterisation of the spreading speed: (Weinberger (1982,...), Zhao(2007,2008,...), Yagisita (2009))..
......
Let f be a bistable, ignition or monostable nonlinearity and assume that Jis such that the assumptions a front exists. Then(i) Variational formulas for KPP type nonlinearities:
c∗ = infλ∈R
1λ
(ˆR
J (−z)eλz dz − 1 + f ′(0))
,
(ii) Linear vs nonlinear determinacy of the minimal speed,(iii) The spreading speed is the minimal speed for the existence of travelling
wave.
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Known Results.Acceleration: (Kot-Medlock (2003), Zhao (2007), Yagisita (2009), Garnier(2011))..
......
Let f be a monostable nonlinearity such that f ′(0) > 0 and assume that J asymmetric fat tailed kernel. Then
(i) c∗ = +∞, and for all c > 0, min|x|<ct
u(t, x) → 1 as t → ∞
(ii) There exists ρ > f ′(0) such that ∀ε ∈ (0, f ′(0)), λ ∈ (0, 1), there existsTε,λ such that for t ≥ Tλ,ε,
min(
J −1(
e−(f ′(0)−ε)t)
∩ R+)
≤ x±λ (t) ≤ max
(J −1 (
e−ρt) ∩ R+)where xλ(t)+ := supx |u(t, x) = λ and xλ(t)− := infx |u(t, x) = λ.
.Remarks :..
......
Position of level sets accelerate like J −1(e−γt)
J(z) ∼ 1|z|α , C0e
(f ′(0)−ε)α
t ≤ |x±λ (t)| ≤ C1e
ρα
t .
J(z) ∼ e−|z|α
, with α < 1, C0t1α ≤ |x±
λ (t)| ≤ C1t1α .
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
A Missing piece in the puzzle
.Remaining open cases..
......
Existence/Non existence of Travelling wave for fat tailed kernel J and amonostable f with some degeneracy at 0 (i.e. lims→0
f (s)sβ = cβ) for some
β > 1)
.Acceleration or not?........ Characterisation of the Acceleration when it occurs when f is degenerated.
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Existence/Non-existence of travelling wave
.Theorem 1 (Alfaro-C. (2016))..
......
Let f ∈ C 1loc(R) be a monostable function such that lim
s→0+
f (s)sβ
≤ Cβ for some
Cβ > 0 and let J such that for α > 2, J(z) ∼ 1zα as z → +∞. Then there
exists c∗ such that for all c ≥ c∗, there exists a monotone travelling wavewith speed c, iff
β − 1 ≥ 1α− 2 .
0
1
2
3
4
0 1 2 3 4 5 6 7 8 9 α
β β = 1 + 1α−2
No Travelling Wave
Travelling Wave
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Acceleration.Theorem 2 (Alfaro-C. (2016))..
......
Let f ∈ C 1loc(R) and J as in the above Theorem and let u(t, x) be the solution
of the Cauchy problem with a front like initial data u0. Assume
β − 1 < 1α− 2 .
Then c∗ = +∞ and limt→∞
xλ(t)t = +∞. Moreover, there exists C0 such that
xλ(t) ≤ C0tβ
(α−1)(β−1) .
0
1
2
3
4
0 1 2 3 4 5 6 7 8 9 α
β β = 1 + 1α−2
Acceleration
Travelling Wave
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Estimates on the position of the Level set.Theorem 3 (Alfaro-C. (2016))..
......
Let f ∈ C 1loc(R) and J as in the above Theorems and let u(t, x) be the solution
of the Cauchy problem with a front like initial data u0.Assume
β − 1 < 1α− 1 .
Then there exists C1 < C0 such that for t large
C1t1
(α−1)(β−1) ≤ xλ(t) ≤ C0t1
(α−1)(β−1) + 1α−1 .
0
1
2
3
4
0 1 2 3 4 5 6 7 8 9 α
β β = 1 + 1α−2
Estimate
Travelling Wave
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Remarks
.
...... Note that 1(α−1)(β−1) + 1
α−1 <1
(α−2)(β−1)
.
...... Note that 1(α−1)(β−1) + 1
α−1 → 1(α−1)(β−1) as α → ∞
.
......The existence of fronts still persists for algebraic kernel, kernel for which
an exponential acceleration is observed when f ′(0) > 0.
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Inside Dynamics
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Inside dynamics of an expanding solution
The expanding population u is made of several neutral fraction υk :
0 x
u(t, x)
υ10 υ2
0
υ30
υ40
υ50
υ60 υ7
0
c→
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
.A coupled system of equations..
......
At t = 0: u0(x) := u(0, x) =∑7
k=1 υk0(x), with υk
0 ≥ 0 for all k ∈ 1, · · · , 7.
The fractions υk only differ by their position and their allele (or their label):∂tυ
k =ˆRJ (x − y)(υk(t, y) − υk(t, x))dy + υk g
(u(t, x)
), t > 0, x ∈ R,
υk(0, x) = υk0(x), x ∈ R.
(3)
g(u) = f (u)/u: the per capita growth rate of each fraction and of the totalpopulation u.
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
The characterization of the inside dynamics of a solution u is madethrough to the generic properties of its fractions.
.Remark:..
......
u and υi , satisfy (3).The uniqueness of the solution of the Cauchy problem =⇒ u(x, t) =∑
i∈I υi(t, x) for all t > 0 and x ∈ Ω.
.Remark:..
......
This Idea of looking at a composite population find its roots in the worksof Vlad et al. 2004 and Hallatschek and Nelson 2008) on the notion of genesurfing. The above mathematical formulation is due to Roques, Garnier andtheir collaborators in 2012.
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
The integro-differential travelling waves
.Thin-tailed kernel: (Bonnefon,C,Garnier, Roques)..
......
• KPP case: If the initial density υ0 of the fraction decreases faster than Uas x → +∞, then
maxx∈[A,+∞)
υ(t, x + ct) → 0 as t → +∞, for all A ∈ R.
• Weak Allee case with large speed: If the speed of the traveing wave is largec ≥ c∗∗ > c∗ then any initial density υ0 of the fraction decreasing faster thanU as x → +∞, satisfies
maxx∈[A,+∞)
υ(t, x + ct) → 0 as t → +∞, for all A ∈ R.
.Remark:..
...... Integro-differential travelling waves have the same inside dynamics aspulled monostable waves;
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Consequences in population genetics
Thin-tailed kernel without Allee effect
Dynamics of the fraction υk with J (x) := Jexp(x) := 12 e−|x|.
Initial structure at time t = 40 at time t = 72
Thin-tailed kernel without Allee effect get a spatial advantage at theforefront .
Integro-differential travelling waves pulled travelling waves;
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Inside dynamics of accelerating solutions
Dispersal kernel: let β > 0,
J (x) = β
π(β2 + x2) for all x ∈ R
Initial data u0: let l > 0,
x0
υ1
0(x) υ
2
0(x)
u0(x)
−l l
.Fat-tailed kernel: (Bonnefon, C, Garnier, Roques)..
......
There exists a time τ > 0 and a constant α > 0 such that
υ1(t, x)u(t, x) ≥ α for all t ≥ τ and x ∈ R.
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Consequences in population genetics
Thin-tailed kernel vs Fat-tailed kernel
Dynamics of the fraction υ1 and υ2.
−2000 −1500 −1000 −500 0 500 1000 1500 2000
0.0
0.2
0.4
0.6
0.8
1.0
−2000 −1500 −1000 −500 0 500 1000 1500 2000
0.0
0.2
0.4
0.6
0.8
1.0
−2000 −1500 −1000 −500 0 500 1000 1500 2000
0.0
0.2
0.4
0.6
0.8
1.0
Initial structure Fat-tailed kernel Thin-tailed kernel
Proportion of υ2 0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 500 1000 1500 2000 2500 3000 3500 4000
0
0.2
0.4
0.6
0.8
1
-2000 -1500 -1000 -500 0 500 1000 1500 2000
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
−500 0 500 1000 1500 2000
0.0
0.2
0.4
0.6
0.8
1.0
(k) t = 0
−500 0 500 1000 1500 2000
0.0
0.2
0.4
0.6
0.8
1.0
(l) t = 20
−500 0 500 1000 1500 2000
0.0
0.2
0.4
0.6
0.8
1.0
(m) t = 36
Figure : Inside dynamics of the positive solution of the Cauchy problem with akernel Jsqrt = e−
√1+|x|/
√1 + |x|, f (s) = s(1 − s) and a Heaviside type initial
data i.e. u0 = 1(−∞,0].
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
The Heterogeneous Situation
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
The non local Reaction diffusion equation
∂tu = J ∗ u − u + f (t, x, u). (4)
J ≥ 0,´R J(z) dz = 1 and
´R J (z)|z| dz < +∞.
f (t, x, s) is a smooth function, f (t, x, 0) = 0..New Problems emerged..
......
Notion of fronts ? Notion of bistability ? The notion of stability of the equilibria ? Shape of the propagation structure ? ”competition” .....
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Some answers in a periodic setting
.Several type of periodicity..
......
time periodic problem space periodic problem time and space periodic problems
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Time periodic problem
J ∈ C 1(R), J ≥ 0, J(0) > 0,´
J = 1 f (t, x, s) = f (t, s) with f (t + T , s) = f (t, s) for all t, s.
.Periodic bistable fronts: Bates-Chen (1999), Fang-Zhao (2015)..
......
Assume J and f is bistable, then there exists a unique c such that thereexists a time periodic function φ(t, x − ct) solution of
∂tφ(t, z) − c∂zφ(t, z) = J ⋆ φ(t, z) − φ(t, z) + f (t, φ(t, z))
Bates-Chen continuity type method. Fang-Zhao Monotone semi-flow construction
.Periodic monostable fronts: Liang-Zhao (2009)..
......
Assume J is thin tailed and f is monostable, Monotone semi-flow theory of Zhao et ale should give existence and char-acterisation of the spreading speed and of periodic fronts.
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
space periodic problem
J ∈ C 1(R), J ≥ 0, J(0) > 0,´
J = 1 f smooth, f (t, x, s) = f (x, s) with f (x + L, s) = f (x, s) for all x, s.
.
......
The notion of planar fronts make no sense in this situation. New notion of front: Pulsating fronts, u(t, x) := φ(x · e + ct, x) such thatφ(s, .) is a periodic function for all s..
......
Spectral problems appears !!!The problem :
J ⋆ ψ − ψ + a(x)ψ = −µψ
may be ill posed in the set C (RN ).
Figure : Simulation of pulsating front moving front the left to the right
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
space periodic problem.Pulsating front: Coville-Davila-Martinez (2012), Shen-Zhang (2012), Zhaoet ale .....
......
Assume J is thin tailed and f is of KPP type. Assume further that for all λthe linearised problem
ˆRN
J (x − y)eλ(y−x).eψ(y) dy − ψ + fu(x, 0)ψ = −µp(λ)ψ in RN .
admit a solution (µp, ψ) with ψ ∈ C (RN ). Then there exists c∗ such that forall c ≥ c∗ there exists a function φ(s, x) solution of
cφs =ˆRN
J (x − y)φ(s + (y − x) · e, y) dy − φ+ f (x, φ) ∀s ∈ R, x ∈ RN
φ(s, x + k) = φ(s, x) ∀s ∈ R, x ∈ RN , k ∈ ZN ,
lims→−∞
φ(s, x) = 0 uniformly in x, lims→∞
φ(s, x) = p(x) uniformly in x,
Coville-Davila-Martinez PDE approach. Shen-Zhang dynamical system approach. Zhao Monotone semi-flow construction
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
(a) t = 0 (b) t = 10 (c) t = 20
(d) t = 40 (e) t = 80 (f) t = 160
Figure : Simulation of the solution of the Cauchy problem when µ0 is notassociated to a eigenfunction function. J is a Gaussian,f (x, s) = a(4x)s(1 − s)where a(x) = 1 −
√|x| on [−1, 1] and is extended by periodicity. Pics are
appearing at the forefront and propagates at a constant speed
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Time and space periodic case
J ∈ C 1(R), J ≥ 0, J(0) > 0,´
J = 1, J is thin tailed f smooth, f (t, x, s) with f (t + T , x + L, s) = f (t, x, s) for all t, x, s, f is
a KPP type.
......
As above the notion of planar fronts make no sense in this situation. New notion of front:“periodic” Pulsating fronts, u(t, x) := φ(x ·e +ct, t, x)such that for all s φ(s, ·, ·) is a periodic function of t and x..
......
More spectral problems appears !!!What is the right notion of eigenvalue for a problem :
∂tψ − J ⋆ ψ + ψ − a(t, x)ψ = µψ.
.Pulsating wave and spreading speed: Shen-Rawal (2012), Zhao et ale ? .....
......
Existence, stability of pulsating wave, existence of a minimal speed Characterisation of the Spreading speed
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
generic nonlinearity
J ∈ C 1(R), J ≥ 0, J(0) > 0,´
J = 1, J is thin tailed f smooth, f (t, x, s) with no a priori structure
.
......
As above the notion of fronts need a right definition. New notion of front:“transition wave” defined by Berestycki and Hamelu(t, x) := φ(t, x) a entire solution which looks like at ± to the stationarysolution.
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
.Some Results..
......
Lim and Zlatos, J compactly supported, f is a time and space“generic” KPP type nonlinearity: Existence of a transition front
Shen-Shen, J is thin tailed, f is a “generic” time bistable or ignitionnonlinearity: Existence, regularity and stability of a transition front
Shen and Shen-Shen, J is thin tailed, f is a “generic” time kppnonlinearity: Existence, regularity and stability of a transition front
.Remark..
......There is no speed associated to these solution Notion of spreading speedto be adapted
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Invasion along a environmental cline
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Biological invasion by adaptation to local conditions
.Generic Context
..
......
A species is usually well adapted to an optimal environment Because of Temperature, light, weather condition, . . ., new
territories are often not optimal for the growth of the invadingpopulation
To colonize this new territories and extend its repartition area, thislocal conditions force the population to have an “adaptation” process
.The case of an environmental cline..
......
The environmental changes follow a gradient of temperature, light,antibiotic, . . .
The expansion of the species is possible through the adaptation ofone or more phenotypic traits
=⇒ Strong relation between the trait value and theposition .
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Example of environmental cline
Studies of phenotypic trait related to budburst, budset andgrowth shows the presence of a latitudinal gradient ((a) Oak, (b)Birch, (c) Pine, (d) Spruce):
O. Savolainen, T. Pyhajarvi et T. Knurr, Annu. Rev. Ecol. Evol. Syst. 2007.
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Clonal Population invading an environmental cline
The Model consideredn(t, x, y): density of population at time t ≥ 0, position x ∈ R, and with aphenotypic trait y ∈ R:
∂tn(t, x, y) − ∂xxn(t, x, y) − ∂yyn(t, x, y) =(rmax − A(y − Bx)2 − 1
K
ˆR
k(y, y′)n(t, x, y′) dy′)
n(t, x, y)
A: Selection intensityB: slope of the clineK : carrying capacity .
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Impact of A and B
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Invasion
.Characterisation of the invasion
..
...... Study of planar wave solutions of the equation: solutions which moveswith a constant speed and constant shape..Travelling wave solution..
......
Solutions of the form ψ(x + ct, z) where c is a speedto be determined and ψ := ψ(ξ, z) is a profile satisfying
∂ξξψ − (B2 + 1)∂zzψ − 2B∂ξzψ − c∂ξψ =(
1 − Az2 −ˆR
k(z, z ′)ψ(ξ, z ′) dz ′)ψ in R2,
with a had hoc behaviour as x → ±∞..Theorem (Alfaro-C-Raoul)..
......
Assume 0 < kmin ≤ k ≤ kmax . There exists a critical speed c∗ > 0 such that,for all c ≥ c∗, there exists (c, ψ) a positive solution of the TW problem.Moreover, no positive bounded front exists if 0 ≤ c < c∗.
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Research IDEas
.Research directions..
......
Many research directions have emerged these last couple of years: Propagation phenomena for the nonlocal Fisher-KPP equation
∂tu(t, x) = ∂xxu + u(1 − k ⋆ u).
Questions: What happens when ∂xx is replaced by nonlocal diffusion Can space heterogeneity prevent acceleration ? Inside dynamics of accelerated solutions Invasion by adaptation process, sexual reproduction model?
. . . . . . .Motivation
. . . . . . . . . . .Homogeneous models
. . . . . . . .Inside Dynamics
. . . . . . . .Heterogeneous model
. . . . . . .Other nonlocal equation Research IDEas
Thanks you for your attention and goodnight